Sense-Making of Critical Incidents with Sentiment Analysis and Data Visualization

In this article, we describe a process for applying sentiment analysis to accelerate the identification of topics and trends in datasets of critical incident responses. Using our work applying the critical incident technique (CIT) at two hospital systems as a case study, we propose a practical methodology to identify trends in narrative data using emotional content assessment and data visualization. An overall framework is provided for engineering managers that is generalizable to other large narrative datasets, such as consumer complaints or survey results, for identifying the patterns in narrative responses more quickly and economically than by manual reviews.